1
$\begingroup$

Recently I came across an information stating that, if we have too many features, the model is most likely to overfit. I not sure why exactly this is happening. I mean, if I don’t use any higher order polynomial equation, but use a lot of features, will the model still overfit?

$\endgroup$
1
$\begingroup$

Yes. The Modell will still overfit if you use too many features. When you use your modell with many predictors for an out-of-sample prediction (e.g. future outcomes or out-of-sample data) you will likely have an increased prediction error.

You should apply a feature selection algorithm or dimensionality reduction to select the most important features.

$\endgroup$
3
  • $\begingroup$ Please tell me how i can extend my answer if you need more detailed clarification. $\endgroup$
    – Ferdi
    Nov 20 '18 at 14:00
  • $\begingroup$ did you mean feature values of test data by predictors? $\endgroup$ Nov 21 '18 at 2:25
  • $\begingroup$ Yes. Feature values of test data both by dependent as well as by independent variables. $\endgroup$
    – Ferdi
    Nov 21 '18 at 8:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.